View Full Version : DNG raw and Cineform some my doubts


Carlo Macchiavello
April 19th, 2014, 03:42 PM
After since a one year of experiment about dng and cineform i have some questions:

I start with conversion of dng sequence by command line and goprostudio
It must be the same process and same result, but it's not real every times...

i have some sequences that are not read, other that read only part of frames (ever last part is missed) from Goprostudio, but DFP2CF.exe read and convert correctly both (luckly for me).

i found that dng converted directly into cineform RAW seems to could collect more noise (also in AD-3) than debayered dng from adobe camera raw and encode cineform in fs2.

my thought is :
DNG decoding -> encoding wavelet cineform Raw -> amplify noise contrast in some way.

More important is that I found i have some issue to recover correctly some dectails on high lights of some images like that picture i post here.

I did some test about debayering, and on high freq dectail seems that Debayer from Adobe camera raw is a bit (very bit) dectailed than CF raw AD3 picture.
To show better i did a test with focus chart, check inner focus point.
i attach files (must be see at 100% zoom).

i tested also dng raw debayering from davinci 10, and it's very bad color and dectails decoding (avoid it everytimes it's a rought debayer way).

For me the excellence quality / efficiency is
Adobe camera raw (where you can normalize picture) + AfterEffects 16bit composition -> cineform mov.
this workflow is faster than encoding raw cineform, since half time on same computer, give me more dectails on shadow and high lights recovered by ACR, more gradable picture, with all advantage of cineform DI.

Now i ask:

1) is my impression to see more noise in debayered picture by cineform than Adobe camera raw?

2) like red code, where new and different debayer system give me during years better decoding result from same shooting, can i hope to see in future new debayer system on cineform?

3) can will see in future a dedicated Debayer algorithm for different camera/sensor like BMC camera? i know that Arriflex have their debayer software, Red same thing, BMC suggest wrong software, the bundled Davinci with BMC, but debayer algorithm is bad and waste too much infos.

4) there is a more dectailed info about recovering highlights dectails on debayered picture by cineform? i read manual but i not be able to recover more info from many highlight sequences in ACR than in firstlight or GoPro studio.

5) Goprostudio give me ability to encode DNG to CF raw with input and output color curve (suggest on CF faq for BMC is Protune), Dpx2Cf.exe not ask me it, this mean some difference? or like i understand is only a preset, but when i set up metadata of CF i can change everything, also encoding and decoding curve?

I think Cineform is the best DI codec that exist, and its unique characteristics make it best solution for DNG and other Raw codec, i promote it on every facility i ever worked, for DI, for stereoscopic DI, and more... and i would like to know how to optimize it with raw.

View image: ACRvs Cineform (http://postimg.org/image/gfzldfhx1/)
View image: Debayer AdobeCamera Raw (http://postimg.org/image/hjjppe2k5/)
View image: Debayer Cineform AD3 (http://postimg.org/image/5vpnuudf9/)

David Newman
April 20th, 2014, 11:57 AM
1) Neither adds noise. ACR has a stronger chroma artifact suppressor, which may be perceived as a change noise.
2) Yes, we also hope to add more debayer options.
3) I don't believe that debayer althorithms should be customized to the sensor. It doesn't work for Red, I expect it doesn't work for Arri also. This suggests that camera vendors know better than you when choosing the look. I'm fan of letting user control everything.
4) There is info, everything is preserved, the trick relates to suppressing the real artifacts of a clipped chroma channel. Use a combination of lowering Exposure and Sat. Clip Pt. with GoPro Studio.
5) DPX2CF doesn't ask but you can set it just like within Studio. Use the switch -ec113 for the default Protune curve.

Carlo Macchiavello
April 21st, 2014, 10:57 AM
this is why i love cineform, you do a simple questions and Master of codec answer to you :-D

1) thanks, i start recently to play with ACR and i found it is quite useful for many color managment, but i don't know exactly what it did in its inner process, i only see the final result.
2) yahoo :-DDD
3) my thought was that it can be optimized on sensor structure, but naturally i also prefer to control the look of my files.
4) i try and i tell you, honestly i not found the right parameters combination to obtain the same dectails.
5) excellent :-D

thanks a lot for faster answer :-D